Re: Move PinBuffer and UnpinBuffer to atomics

On Thu, Apr 14, 2016 at 5:35 AM, Andres Freund <andres@anarazel.de> wrote:

On 2016-04-14 07:59:07 +0530, Amit Kapila wrote:
> What you want to see by prewarming?

Prewarming appears to greatly reduce the per-run variance on that
machine, making it a lot easier to get meaningful results.  Thus it'd
make it easier to compare pre/post padding numbers.

> Will it have safe effect, if the tests are run for 10 or 15 mins
> rather than 5 mins?

s/safe/same/? If so, no, I doesn't look that way. The order of buffers
appears to play a large role; and it won't change in a memory resident
workload over one run.

I've tried to run read-only benchmark of pad_pgxact_v1.patch on 4x18 Intel machine.  The results are following.

clients no-pad  pad

1       12997   13381

2       26728   25645

4       52539   51738

8       103785  102337

10      132606  126174

20      255844  252143

30      371359  357629

40      450429  443053

50      497705  488250

60      564385  564877

70      718860  725149

80      934170  931218

90      1152961 1146498

100     1240055 1300369

110     1207727 1375706

120     1167681 1417032

130     1120891 1448408

140     1085904 1449027

150     1022160 1437545

160     976487  1441720

170     978120  1435848

180     953843  1414925

snapshot_too_old patch was reverted in the both cases.

On high number of clients padding gives very significant benefit.